Re: computer-go: Programming languages

[David Fotland <fotland@xxxxxxxxxxxxxxx>]

At 07:34 PM 9/3/00 +0200, Heikki Levanto wrote:
>William Harold Newman <william.newman@xxxxxxxxxxxxxxxxx> wrote:
>
>> Also incidentally, the Gamesman's Toolkit was the thing which
>> convinced me to switch my Go-playing program from C++ to Common Lisp.
>> Doing the Gamesman's Toolkit thing in Lisp is still messy, but it's
>> substantially less messy than doing it in C++ (or C!). I took this as
>> strong evidence that some fundamental problems in Go would be easier
>> to analyze in Lisp (or, probably, in a functional language like ML)
>> than in C++. I still think that's correct, but man, oh man, they're
>> still hard problems. I've been working on them a long time since then,
>> and I still haven't been able to turn my analysis into something which
>> plays a decent game of Go.
>
>Yes, many of the languages in use have their own strong points. In my life I
>have started to implement go in at least: Assembler, C, basic, Pascal,
>Prolog, Java, Perl, and some others I do not remember. They all seemed to
>have their advantages and drawbacks. My conclusion (so far) is to use what
>ever language I feel most comfortable with. Some other language may have
>some benefits I miss, but will these compensate for the trouble of learning
>a new way of doing things?  Maybe, but can I know that in advance?

No matter how much fancy analysis goes into "understanding" the position
or high level strategic analysis, a program is only as good as its reading
ability.  So the tactician at least should be written in something that
executes fast :)

-David